Don't send extra_float_digits for PostgreSQL 12+ by default

[NKame]

Describe the issue
Connecting through PgBouncer with a default configuration fails.

Driver Version?
All supported.

Java Version?
All supported.

OS Version?
All supported.

PostgreSQL Version?

= 12

To Reproduce
Connect throught PgBouncer. Fails with unsupported extra_float_digits parameter.

Expected behaviour
Connected to the database.

Context
extra_float_digits has been introduced in PostgreSQL 8.4 to play around the apparent precision of the lossy datatype "float". At first to increase the precision sent to the client, then since version 12, to reduce it. In pgJDBC, if you want the "full" precision with a server 12+, you must override this parameter since it is forced in the driver to the value 3 if the server is 9+, that means you have to send it twice to the server. Furthermore, by default it's not authorized by default with PgBouncer: "the internet" answer is to change the server, whereas it doesn't make really sense to send it at all for PostgreSQL 12+. (And the question of forcing a default for the lossy datatype should be asked... in my view it should be an explicit choice from the application owner).

[jorsol]

PgBouncer should be configured with ignore_startup_parameters = extra_float_digits
https://www.pgbouncer.org/config.html#ignore_startup_parameters

[NKame]

I address precisely this thing in the context section. But that doesn't answer the secondary main concern: why reduce arbitrary the float precision? Why should we send 2 instructions to get the native precision when none is required at all?

[jorsol]

https://www.postgresql.org/docs/17/datatype-numeric.html#DATATYPE-FLOAT

Applications that wanted precise values have historically had to set extra_float_digits to 3 to obtain them. For maximum compatibility between versions, they should continue to do so.

[NKame]

Same link, but the information is the the same since 2019.
By default, floating point values are output in text form in their shortest precise decimal representation; the decimal value produced is closer to the true stored binary value than to any other value representable in the same binary precision.
For compatibility with output generated by older versions of PostgreSQL, and to allow the output precision to be reduced, the [extra_float_digits](https://www.postgresql.org/docs/17/runtime-config-client.html#GUC-EXTRA-FLOAT-DIGITS) parameter can be used to select rounded decimal output instead.
And reading furthermore
The meaning of this parameter, and its default value, changed in PostgreSQL 12
and even further
If the value is 1 (the default) or above, float values are output in shortest-precise format; see [Section 8.1.3](https://www.postgresql.org/docs/17/datatype-numeric.html#DATATYPE-FLOAT). The actual number of digits generated depends only on the value being output, not on the value of this parameter.

So I was wrong assuming you needed to reset it to get full precision. Since version 12 it is not needed anymore, so the breaking of PgBouncer default conf on PostgreSQL 12+ is even more meaningless.

[ecki]

It’s fixed in #3490 #3491 now